基于STSCKF和改进型χ2检验的模拟电路故障辨识  

Analog circuit fault identification based on STSCKF and improved χ2 test

作　　者：李小民[1] 杜占龙[1] 郑宗贵 张国荣 毛琼[1] 

机构地区：[1]军械工程学院无人机工程系,石家庄050003 [2]第二炮兵研究院,北京100085 [3]厦门警备区,厦门361003

出　　处：《仪器仪表学报》2014年第10期2248-2255,共8页Chinese Journal of Scientific Instrument

基　　金：总装院校科技创新工程(ZYX12080008)项目

摘　　要：针对模拟电路中不可直接测量故障元件定位和参数估计问题,提出一种基于强跟踪平方根容积卡尔曼滤波(STSCKF)和χ2检验的故障辨识算法。根据强跟踪滤波理论,将χ2检验引入到强跟踪SCKF的残差异常检测中,解决了强跟踪SCKF缺乏故障定位能力的问题。改进传统的χ2检验法,增强其对缓变故障的检测灵敏性。首先利用改进型χ2检验法检测不同故障模型的STSCKF残差输出,确定故障元件,然后采用STSCKF对故障元件参数进行估计。有源滤波电路仿真模型和某型无人机发射机实际故障数据的实验结果表明,相比于传统的χ2检验法,改进型χ2检验能更早地定位故障参数,STSCKF估计误差分别低于强跟踪无迹卡尔曼滤波(STUKF)估计误差的90%和普通SCKF估计误差的5%。Aiming at the unmeasured faulty component location and parameter estimation issue of analog circuit, a fault identification al- gorithm based on strong tracking square-root cubature Kalman filter （STSCKF） and χ2test is proposed. According to strong tracking filter theory, the χ2test is introduced to residual abnormal checking of strong tracking SCKF, which solves the problem that strong tracking χ 2 SCKF lacks the ability to locate fault. The tradmonal χ2 test is improved to increase its detecting sensitivity to slow changing fault. Firstly, the improved X test is utilized to locate the faulty component via detecting the STSCKF residual outputs generated with different faulty models. Then, the faulty component parameter is estimated with STSCKF. The experiment results of active filter circuit simulation model and the actual failure data from a certain unmanned aerial vehicle （UAV） transmitter show that compared with traditional χ2 test χ 2 method, the improved X test can locate the faulty parameter earlier. The STSCKF estimation error is less than 90% of the STUKF （ strong tracking unscented Kalman filter） estimation error and 5% of conventional SCKF estimation error.

关 键 词：强跟踪滤波 状态和参数联合估计 平方根容积卡尔曼滤波(SCKF) 残差检验 

分 类 号：TH165.3[机械工程—机械制造及自动化]